►
From YouTube: SimPEG meeting March 8
Description
SimPEG meeting on March 8
B
C
C
D
A
A
A
D
A
But
it's
sort
of
unclear
where
things
have
in
common
parviz
is
hot.
Oh,
so
you
actually
have
at
least
a
little
segment
over
a
net
her
page
on
each
application.
I
think
eventually
departed.
You
Phoenix,
but
that
might
be
a
bit.
Much
for
now
is
look
if
you
should
get
me
to
have
like
when
you
look
at
this
new
paper,
so
you
can
see
electromagnetic
tonight
crowd
for
you
see.
D
D
A
A
A
Yeah
and
then
one
of
the
things
in
terms
of
examples
and
if
Rowan's
able
to
jump
in
he
can
fill
us
in
a
little
bit
on
this
as
well,
is
giving
better
attribution
both
in
terms
of
the
packages
and
the
examples.
So
I
think
that'll
come
like
packages,
and
these
my
take
on
it
would
be
the
first
place
to
do.
That
would
be
these
sorta
like
at
the
top
of
the
documentation.
A
So
when
you
hit
mag,
for
instance,
like
prime
contact
is
dawn
and
just
make
sure
that
those
people
are
highlighted
and
then
also
in
examples,
what
I
think
it'd
be
kind
of
cool
is
to
do
like
what
we're
doing
in
EMG
osai.
With
the
case
histories
is
like,
you
have
a
name,
you
can
click
on
the
name,
and
it
takes
you
to
a
page
with
all
of
the
people
listed,
and
your
am
not
sure
if
you've
seen
that.
A
See
if
it
works,
so
we
have
a
page,
and
this
is
auto-generated
just
of
people
who
have
contributed
and
then,
if
you
go
to
case
histories
for
instance,
and
pick
one
like
Mount
aiza,
then
the
the
contributors
they
are
listed.
So
it
just
makes
sure
that,
like
people
are
highlighted-
and
another
thing
we
can
do
with
a
lot
of
the
examples-
is
make
sure
that
there's
a
citation
included.
If
it's
like
an
example
from
a
paper
or
something
like
that
to
make
sure
it's
clear
how
to
attribute
work.
H
A
For
ya
and
then
joint
inversions
and
I'll,
let
you
talk
a
bit
about
this
later
on
to
dom
so
the
composite
objective
functions
are
like
in
a
place
where
you
can
play
with
them.
I
think
before
getting
it
into
dev.
I
would
certainly
appreciate,
like
some
thoughts
and
a
review
from
you
from
Rowan
and
from
Joe,
because
I
know
you
guys
have
you
sort
of
been
involved
a
little
bit
in
then
yeah,
so
just
to
make
sure
it's
at
least
more
clear
enough
to
sort
of
have
other
people
jumping
in
before
yeah
yeah
yeah.
F
A
B
I
A
So
dis,
Critias
ation
I'm,
really
excited
to
see
how
quickly
things
have
been
moving
here.
So
this
is
done
now
like
we
pulled
it
out.
It's
in
dis,
critize
now
Joe
is
jumping
in
and
improving
speeds
on
oak
tree,
which
is
really
exciting.
So
I
haven't
heard
much
from
him
on
that
recently,
but
I
know
he's
been
playing
around
the
fun.
Can.
I
A
A
A
B
A
A
A
Make
a
mess
but
I
think
like
for
the
casing
for
a
primary
secondary
type
approach.
It's
actually
like
pretty
solid,
because
then
you,
you
simulate
a
simple
background,
get
the
casing
right
and
then
do
3d.
So
the
main
piece
that
I
need
to
do.
That,
though,
is
interpolation
now
between
a
3d
so
much
and
a
3d
Cartesian.
So
we
don't
have
that
right.
Now,
oh
yeah.
A
F
A
F
A
F
K
F
No
not
sure
okay,
we
could
it
be
definitely
like.
I
have
some
more
option
ality
for
next
politician
or
should
I
tell
monster
yeah.
So
that's
not
our
time.
So
all
of
the
new
national
pastime
options
you
can,
sir
I
do
not
thanks
for
called
stuff.
So
there
are
a
few
things
to
be
done:
I
guess
more,
like
no
I
have
swimming
population
yeah.
A
F
A
Yeah
we've
been
walking
through
the
the
road
map
here
yeah,
so
Joe
has
been
improving
a
lot
with
three
in
terms
of
speed,
so
that
helps
with
operators.
But
we
still
need
you
pills
and
things
like
that
are
actually
creating
meshes
and
just
some
examples
around
that,
because
there's
a
lot
of
good
stuff
in
sim
paid
for
doing
that.
But
it's
not
necessarily
easy
to
find
right
now:
you're,
improving
variation,
harty,
the
operator
creation,
specifically
so
he's
written
a
bunch
of
stuff
like
he's,
taken
it
down
to
C++.
L
A
C
L
A
A
A
F
Kurt
were
there
back
then
that
you
can
have
some
sort
of
utilities
like
a
ubc
types
of
facilities,
that
we've
got
the
topography,
then
now
we
can,
if
you
can
generate
the
fine,
this
could
well
discretize
Nashville,
both
up
tree
and
purple
in
your
mesh.
I
think
that'll
be
a
good
start
to
have
then
actually
useful
people
can
use
it,
at
least
for
DC
case
and.
F
L
L
A
B
A
L
L
No,
I
think
yet
mainly
creating
examples,
especially
with
the
octree
side
of
things,
contrasting
that
with
curvilinear
need
to
talk
to
Joe
a
bet
in
theory,
the
speed
and
still
making
sure
that
we
can
edit
it
so
that
it's
not
off
skated
sort
of
into
something
that
only
Joe
can
touch
and
then
also
the
more
boundary
conditions
and
sort
of
working
on
some
of
the
stuff.
That
song
is
done.
Soggy
and
Mike
I
think
I've
done
on
the
in
the
DC
problem
and
making
that
a
bit
more
accessible.
F
A
C
F
Well,
I
think
my
feeling
when
I
was
visiting
this
station
like
it
at
South
Asian
countries,
and
it
is
really
big-
it's
their
main
application
and
then
their
lack
of
computational
capability
either
like
both
in
simulation
and
inversion.
So
I
think
they're,
mostly
using
whoa,
is
that
we
are
child
you're,
a
high
school
now
poor.
Imagine
so,
but
it
seems
like
it
pretty
hard
for
them
to
run
the
for
moly
I.
Don't
know
why
let
thee
so
yeah
I
think
opening,
like
those
people
like
MGMT
coats,
integ
empty
code
will
be
pretty
useful.
Yep.
G
A
So
the
next
couple
items
on
the
list-
minimize
that's
what
it
needs
to
be
thought
out
again
and
hopefully
get
like
compatible,
at
least
for
sundry,
optimization
in
some
cases,
it's
probably
worth
us
to
bring
our
around
my
vision,
but
would
be
nice
have
a
possibility
because
we've
not
currently
I
had
better
message
what's
wrong.
Do
you
have
comments
on
the
after
party?
Oh.
L
Now,
that's
all
I
one
function
call,
so
we
can
do
that
in
a
decent
way,
but
some
of
that
stuff
will
need
to
get
be
get
the
rethought
if
we're
breaking
them
up
into
other
functions
and
like
knowing,
when
the
model's
changed,
knowing
when
we
can
recompute
or
whatever
fields
and
stuff
like
that,
so
just
making
sure
that
we
don't
lose
the
efficiencies
that
we've
gained
when
we
break
this
out
potentially,
and
so
it's
instinct,
it's
a
little
bit
intricate.
That
piece
and.
L
Ya
know,
I
think
that's
absolutely
true
like
this.
This
isn't
this
is
not
tackled
at
all
in
isolation
like
it
touches
basically
all
pieces,
yeah
yeah,
it's
like
right
at
the
end
of
our
stack,
so
it
needs
to
get
refactored
back
into
some
of
the
other
pieces,
especially
especially
with
the
parallelization
in
the
well.
What
we
suspect
that
is,
the
global
problem,
yeah.
A
A
A
A
A
better
job
same
things
as
well
like
specifically
pulling
out
parametric
stuff,
because
there's
a
lot
that
can
be
leveraged
between
the
definition
of
the
arc
tan
function
and
derivative
will
be
using
all
of
their
magic
examples.
So,
like
yeah,
getting
a
little
more
consistent
to
see
enough,
as
well
as
for
things
like
our
chief
law
and
empirical
relations.
Some
of
those
we
should
put
to
geo
Anna.
So
clearly
the
analytic
functions
in
is
handy
to
have
for
Kusile,
but
then
be
able
to
translate
that
to
a
packet
in
the
mappings
project.
L
L
Because
I
think
some
of
the
other
stuff
Lindsay's
been
working
on
is
the
name
is
escaping
me,
intermediate
step,
effective
medium
theory.
Yet
that's
not
something
that
really
needs
to
reside
in
simpang
and
it's
sort
of
expanding
the
scope
there,
but
it
can
potentially
reside
in
something
like
geo,
Emma
and
then
wrapped
up
into
mappings
and
then
put
into
simple
so
everything's
working
as
a
sort
of
mapping.
There.
C
L
C
L
L
So
I
think
the
the
big
piece
here
is
actually
getting
more
explicit
about
what
we
actually
want
to
save
when
we
say
save
as
in
peg
class
or
try
to
pickle
it
up
or
something
like
that,
and
that
is
starting
to
happen
through
some
of
this
work
with
properties
in
terms
of
documenting
things.
As
as
we
go
and
I
can
point
you
to
a
talk
on
that
and
maybe
give
another
you're
Lindsay.
A
Yeah
I
thought
suppose
be
pretty
to
the
idea
of
a
command
line
interface
or
making
it
easier
to
run
fascism,
simulations
slow
and
tasting
stuff.
I've
done
like
the
way
I
like
to
work,
is
to
set
up
the
mesh
and
the
stimulation
in
the
notebook,
so
I
see
what
I'm
doing,
but
then
I
need
to
get
over
the
posterior
run.
It
so
I've
been
doing
properties
to
basic.
Please
just
save
everything
that
I
meet
you
and
then
going
we
create
whatever
is
he
came
by
the
simulation?
A
A
To
recreate
things
in
a
match,
oh
it's!
Instead
of
writing
the
entire
script
over
and
over
again
I
just
have
look.
Two
files
show
you
very
briefly
like
this
and
I
mean
idea
so
like
this
is
how
I
want
to
run.
Simulations
is
basically
generates
my
parameters,
files,
which
is
a
JSON
file
that
says
casing.
Is
this
comment
to
be
the
statement?
Whoever
I'm.
G
A
L
L
Lowers
the
barrier
to
entry
84
a
lot
of
people
who
are
trying
to
use
synthetic
on
the
outside,
so
I
know:
Dom
has
been
working
with
some
collaborators
and
that's
more
of
a
style
that
they're
familiar
with
and
like
a
first
step
into
this
ecosystem
of
actually
giving
people
like
executables
from
the
command
line
and
saying
edit,
this
JSON
file
or
config
file
or
whatever,
and
then
just
run
the
code
instead
of
giving
them
a
few
scripts
with
all
of
the
all
over
the
pieces
in
it.
Yeah.
F
F
A
This
is
all
on
chasing
research
repository,
so
I'll
show
you
quickly
this
so
basically
have
this
like
I've
got
a
mesh
generator
that
takes
minutes
tells
lies
x1,
which
is
like
close
to
the
casing.
Part
x,
2
is
outside,
and
so
all
I
do
is
basically
creating
my
match
generator
and
then
save
all
of
these.
This
is
all
that.
F
A
B
A
A
K
L
A
L
Yeah,
I
think
the
other
times
dope,
but
don't
lay
today
on
talking
about
research
practices
and
like
lindsay.
You
can
talk
about
maybe
how
you're
setting
up
a
research
repository
to
push
on
like
casing
research
on
them
talking
also
about
how
were
like
writing
my
tech
document
and
stuff
like
that
collaboratively
and
pulling
things
I'm
as
well.
Yeah.
G
No
I
like
that
I
played
yes,
okay,
cool
back
she's,
also
just
keep
in
mind
kind
of
what
we
consider
to
be
best
practice,
ism
and
because
I
mean
I.
Think
like
I
mean
the
notebook
has
a
lot
of
capabilities
of
sharing
and
connecting
two
clusters
and
running
them.
I
mean
running
remotely.
So
that
might
be
something
we
should
also
consider.
Yeah.
G
L
G
Doesn't
really
matter,
but
it
needs
to
communicate
in
the
same
way
on
both
at
exactly
that's
the
important
thing.
Sorry,
but
I
mean
I've
used
this
parallel
remote
or
parallel
functioning,
the
notebook
in
it.
Pretty
I
mean
it's
the
notebook,
it's
so
simple
and
sleek.
So
it
will
be
very
interesting.
Try
it
and
in
trying
to
incorporate
that
into
that
impact.
The.
A
L
A
A
L
Yeah
I
think
especially
like
Oh.
Some
of
these
last
ones
are
like
sketching
at
the
future
and
like
sort
of
shows
some
issues
and
things
that
we
know
that
we
should
be
doing
but
like
higher
up
in
the
document,
there
is
like
that's
I
would
say
where
we're
working
on
immediately
and
inland
here.
Your
future.
J
L
J
It's
just
a
single
parent
organizations.
We
next
see
sale,
yeah
and
it's
fine,
just
one
layer.
Listen
as
you
start
using
wires
and
setting
them
then
call
becomes
a
second
pillow,
and
so
every
time
you
want
question
directed
us
change,
the
authors
changed
the
weights
whatever.
Then
you
need
a
check
on
on
on
them
and
Google
one
or
you
inside
right.
J
A
J
A
D
D
H
I
I
J
J
J
D
J
J
J
F
J
J
Yeah
yeah
he's
he's
that
yeah
first
first
coming
our
way
exactly
and
and
so
she
published
a
3d
go
CAD
model
which
you'll
sign
it
she'll
send
to
me
soon.
That's
the
seismic
cube
and
then
there's
another
guy
who
did
their
ram
now.
So
that's
why
I'm
interested
into
it,
and
it
goes.
They
studied
remnants
its
information
with
I.
J
E
B
D
J
J
I
A
J
E
H
I
J
J
J
And
so
when
we
were
ready
for
joining
version,
that's
going
to
be
our
our
beast,
lady
yeah
and
so
I'm
not
ready
to
Ivan,
pushing
anything
on
the
cloud,
but
I'm
really
working
hard
on
the
on
the
combo
of
two
functions
rather
realizations
work
so
should
be
done
this
week.
Hopefully
comment
striking
up
somebody's.
A
To
say
the
potential
action
items
that
I
would
love
to
hear.
One
is,
if
you
have
thoughts
on
this
road
map,
will
sort
of
finalized
soon
with
that
and
walk
five
finalized
for
now
and
get
it
posted
on
the
superb
website,
oh
and
it
that
so
that
would
be
great.
It's
good
all
right
well
see.
Is
there
later.